adam.oliner.net

I am excited to announce the completion of my Master’s of Engineering thesis, titled Cooperative Checkpointing for Supercomputing Systems. This thesis is the result of work done both at IBM as part of the BlueGene System Software Group and at MIT under the advisement of Larry Rudolph. It was my honor and privilege to work with them, and I am grateful for everything they did to make this thesis possible. I acknowledged a bunch of people in the document, but I hardly get started. Putting just my name on the byline seems almost dishonest; thank you all, again.

Still another bit of good news: in addition to my fellowship with the DOE, I have also been selected as an Honorary Stanford Graduate Fellow.

To reflect all these changes, I’ve done some spring cleaning on the site. The About page is now a bit more accurate, though perhaps premature. I updated the Writings page to include copies of all my publications, including my thesis. Furthermore, I posted a short novella I wrote a couple of years ago, called Glitterbug, that somehow never made it online.

Enjoy!

This is the last scene in a story that started on September 4, 2003, when a chance meeting with Martin Rinard started my journey to begin a Ph.D. It ends today, with my acceptance of one school’s offer of admission, and all the melodrama that I feel is appropriate for such an occasion. In a comment to that above entry, Sean prophetically noted that not all metastabilities have time to decide on their own; life, he said, is full of timeouts. Today is one of them.

My blog charted the course of this story. I mentioned my rushed application to and subsequent rejection from MIT’s PhD program, the next year’s round of applications, their responses, and, penultimately, the visit weekends. The decision process was, for me, tortuous. Patrick Winston noted that people, given two options, will generally take one of two approaches to making a selection: (1) initially see one as better, then rationalize why the other isn’t so bad, until they are equal again or (2) initially see one as better, and continue to reinforce that belief. I, he correctly observed, am a Type 1.

Thus, in addition to much soul-searching and introspection, I also talked with lots of people. I called professors at the schools. I met with professors at MIT who were alums of these schools. I talked with professors, some of whom didn’t even really know me, nor the schools. I took email polls. I solicited IMs. I chatted with friends, and with office mates. I browsed the websites of research groups. I skimmed their publications and recent Ph.D. dissertations. I talked with my family. I weighed pros and cons. I tried the different decisions on for size. I waffled. I paced. I even, God help me, flipped coins to see which side disappointed me less. It was one of the hardest choices I have ever had to make, which is strange because the worst case scenario was still really, really good. “The hardest decisions to make are often the ones that matter the least.” At the end of the day, I had to choose. It was time. And I knew.

This story ends and the next begins like this: on April 15, 2005, I accepted the offer of admission to the Computer Science Ph.D. program from Stanford University.

Thanks to everyone who helped me make my decision, no matter your input nor the form it took. For the record, advice included “X rules and Y drools,” a discussion of the early American settlers (and an obtuse reference to the Oregon Trail), a note on the poop-eating and face-licking habits of dogs, and, finally, “Kill yourself.” You all put in way more effort than I could ever have asked for, and I’m powerfully grateful. Thank you.

Although I wasn’t quite dead in Denver, I sure felt close, at times. Illness is rare for me, with the last case (that I can remember) being the Venezuelan Black Plague. That same disease, brought from distant lands by Daniel “Space Monkey” Loreto, put Carl out of commission for more than a week. When I get sick, I don’t mess around.

On the flight from Boston to Chicago, I sat next to a Mormon gentleman, who also happened to be a computer engineer. He was a perfectly nice guy, but near the end of the flight he began talking about religion. He told me a story about his priest(?), who was asking a member of his church how long it would take him to calculate the intersection point of two objects moving in the air. The member responded that it would take him a while to determine the equations, measure factors like gravity and air density in the area, and so on. The priest then tossed his keys at the man, who caught it in mid-air. This, my fellow airline passenger asserted, was a profound testimony to the miracle of God’s creation.

Not being able to help myself, I responded that the comparison seemed a little rigged, as preparation time was included in one but not the other. The human body, I noted, took millions of years to evolve the physical mechanisms necessary to perform that little miracle. Even if you just start counting from birth, it takes several years of learning and practice; go throw some keys at an infant if you need proof that there’s some prep time involved. Once you’ve written the program, I concluded, a computer could calculate the midair collision of billions of pairs of objects in a couple of seconds. Most people I know can’t juggle more than three objects.

There was a moment of silence as the Mormon gentleman pondered my point. Then, seemingly undiscouraged, he asked if I had read the Book of Mormon. “No,” I conceded, “I haven’t. But I’m fine, thanks.” To a normal person, this is a clear signal of disinterest in a line of conversation. Instead, he began to describe to me the similarities between his religious text and the Bible, and then flowed non sequitur into stories about John the Baptist appearing in Pennsylvania in 1829. Those who know me might imagine the difficulty with which I held my tongue. The temptation to draw comparisons between his fantastic stories and, say, Greek mythology were strong. But I resisted, because we were almost at the gate, and all I would be doing is angering a stranger. I just nodded, avoided eye contact, and went “hmm”. He suggested I visit mormon.org, gave me a card, and we disembarked from the plane.

The weekend I spent at UIUC coincided with the Fighting Illinis’ big Final Four game. On the shuttle from the airport, we drove through Champaign, passing many bars and restaurants near campus. From the door of each establishment poured hundreds of bright orange bodies. Students, clad in their school colors, preparing for the game by starting to drink at two in the afternoon. Illinois, like Stanford and CMU, was nice. Smart, friendly people. Lots of good stuff. The hotel placed me in a smoking room, despite my voiced preference otherwise, and I was too tired to do anything about it. This triggered some allergies, which distracted my immune system, which was, finally, taken over by the cruel illness I mentioned earlier.

My flight to Denver was spent in better company: a clever six-and-a-half year old black girl. She watched the modern live-action version of Cinderella, which, she confided to me, was her favorite movie. She had a doll named Heather, who the flight attendant enjoyed pretending was a real child. “Make sure you hold onto her real tight,” the woman said, “You don’t want her to get hurt.” The little girl was amused that the woman believed her doll to be a real person, and she and I spent the flight pretending that the stewardess was a little crazy. As we descended toward DIA, my conversation with her (such as it was) turned to the subject of weather. “Do you know what clouds are?” I asked her. She shook her head. “They’re water,” I explained. “And when it rains, that’s just the clouds getting too heavy and falling out of the sky.” She stared at me, mouth parted slightly. I continued, “And fog is just a cloud that’s too low. So when you walk through the fog, you’re walking through a cloud.” The little girl gasped softly, her wonder-filled eyes growing wide, images of Cinderella walking through the low-hanging clouds flashing in her mind.

“Catching keys,” I thought, “that’s what passes for a miracle?”

The conference was in a nice hotel near Denver, and I had what they called a mountain view. It was really a parking lot view, but, sure enough, the snow-covered Rockies loomed in the distance. My talk went well, but I got rejected from a couple of fellowships. Oh well. I arranged with IBM to remain in Boston for the summer, doing research with Larry and coding an implementation of my Master’s thesis (cooperative checkpointing) for BlueGene. My decision for graduate school is due this Friday. I am still struggling. This weekend I will blog again, posting my choice. Either way, it seems my days as an MIT student are numbered. Before landing in Logan, we flew over Boston and Cambridge in the twilight. I could pick out the Great Dome, the Green Building, Sydney and Pacific, the Stata Center, the Pru, and the brilliantly lit Citgo sign. It was beautiful, but I was struck by a sudden sadness. I realized why: soon, I’ll have to say goodbye.

Incidentally, don’t ever throw your keys at an infant. It was just a thought experiment. Sicko.

My weekend began at 2:44 AM on Friday, when I was awakened by my cell phone. I had hastily instructed Orbitz to call me 3 hours before my flight departure time, without consideration for when, precisely, that would be. Begrudgingly, I finished packing and got ready; my cab was due to arrive at 4. Before the sun had risen, I was in the air, headed to San Francisco. To Palo Alto, really. It was Stanford’s visit weekend.

The path I walked as I exited SFO took the form of a loop; I walked out, walked right back in again, and stuffed my jacket into my suitcase. It was the hottest March 11 in San Francisco history. The sky was, well, blue. At the Sheraton in Palo Alto, where Stanford put me, there were people swimming in the pool. As these two screenshots clearly illustrate, the weather difference between Boston and SF was nothing short of comical. I immediately changed into shorts and a T-shirt, donned my sunglasses, and meandered to campus.

There were events planned throughout the weekend, including campus and housing tours. On Friday night, a group of us hopped on the BART and went into San Francisco for dinner (at Ramblas Tapas). We followed that up with a walk through Chinatown, coffee in Little Italy, and hiking the streets. Saturday was mostly business: information sessions, seminars on the various research groups, and meetings with professors. I spent the evening relaxing with some other admits and grad students. We played Taboo, and I later joined forces with the talented Finale to trounce some non-MIT folks at Foosball.

Unfortunately, I took most of the pictures on Sunday morning, before returning to MIT. This means you don’t really get the same sense of student life as I witnessed on Friday. For example, this fountain was filled with people in bathing suits and inner tubes, catching some sun. This field was populated by students lounging on beach blankets and playing frisbee. Although there are better factors than weather based on which to pick a Ph.D. program, Stanford certainly made a compelling case in that regard.

This is only the beginning, however. Next weekend I’ll be at Carnegie-Mellon, in Pittsburgh. I won’t be packing my shorts.

I am approximately midway through my notification deadline checklist, and with mostly good news. I have been accepted to the Ph.D. programs at Stanford, Carnegie-Mellon (CMU), and the University of Illinois at Urbana-Champaign (UIUC). MIT and Berkeley are both still thinking about it. Good for them. I’d hate for them to rush and do something foolish, like reject me. Once those have been nailed down, we’ll return to this topic.

Schools aside, there were also papers and fellowships. The Hertz Foundation has deemed me unfit to receive their money, which bothers me not-at-all. There are still the NSF, DOD, and DOE fellowships for which I am anxiously awaiting responses. For their part, UIUC has offered me a generous departmental fellowship package, in the hopes of wooing me to their fine school.

I’ve also been fortunate paper-wise. Although my initial submission to IPDPS 2005 was rejected, a subsequent submission to a workshop in that same conference was enthusiastically accepted, thus raising questions about the consistency of the reviews. Those aside, I will be presenting that paper in Denver this April. I’ve presented a paper at IPDPS before, so my giddiness is somewhat tempered. This is especially true in light of my other paper acceptance, at DSN 2005 in Yokohama, Japan. Not only does that paper mark my first co-authorship with an MIT professor (my advisor), it will also be my first time outside of North America.

Starting in a couple of weekends, I will begin the process of visiting schools. I’ll be flying to San Francisco to visit Stanford, Pittsburgh for CMU, and Chicago for UIUC. The Illinois visit overlaps with the start of IPDPS, so I will be flying directly from there to Denver to give my talk. Plus Japan in June, and maybe a family trip to San Diego in August. Frequent flyer miles will be pouring forcefully from every orifice. Don’t try to picture that, you’ll regret it.

In the mean time, I’m working in my office in CSAIL. Aside from revising papers and working on my thesis, I’m also involved with a class project (with Daniel and Erik) to make MIT an official supercomputer site. That means taking a collection of computing clusters and getting them to solve an enormous system of linear equations really, really fast. If we can run this benchmark and get an average of about one trillion floating point operations per second, we should make the Top500 List. Which would be pimp; it would mean I had put MIT on the list and helped put IBM at number one. Finally, my duties as a TA have earned me the nickname Adam the Merciless.

And that was the short version!